The Story of Dhravya Shah | 20yo raised $3M to build SuperMemory

Today's episode eight, and we have got Kavya Shah, the man of the hour. Twenty years old, raised three million dollars, built memory for AI. I mean, you worked at MemZero, you worked at Cloudflare.

Entire like Indian news push has been really funny to me because I I never got into IIT. I never dropped out of IIT, and then the acquisition happened somewhere at the end of eleventh grade. I was like, dude, now why do I even need to go to IIT? I see all these IITs. Salaries and I'm getting it right now. Like, what's the point? When I built the first version of Any Context, one of my other random products called dumb.place, it got attacked. So I had to reverse engineer Cloudflare to save myself, and then I accidentally got a job at Cloudflare. Came into ASU where I studied with this mindset of I want to build a lot, and so like no girls, no parties, no chill. You know, I just want to build. So I I was just building and She saw my failure tweet, and then she she was she was walking past me, and she came and she was like, "Dude, I saw your tweet. You wanna go for a walk?" I was like literally I had run out of all my money from acquisition by this time. I I had borrowed a thousand dollars from my friend because I was in the U.S. Okay. I had borrowed a thousand dollars from my friend and eight hundred dollars from my girlfriend. So that happened, and then this random guy comes to me, and he's like, "Dude, I have the domain sophomemory dot a." Do you just want it?

Hello, ladies and gentlemen. Welcome back to Ground Zero AI Talks. Today's episode eight, and we have got Dhavya Shah, the man of the hour. Hi, hi. Welcome, Dhavya. How are you feeling?

I'm feeling great. Thanks for inviting me, Himanshu.

Pleasure to have you here. So, Dhavya, let's start from the from the news itself. Twenty years old. Raise three million dollars, build memory for AI. Dude, when you hear the sentence back, how does it feel?

Yeah. I mean, I could never have imagined myself going through the venture-funded path. But now that I'm here, it's kind of great. It's also a lot of responsibility, of course, because now I have to you know make it big in like the next two three years before I turn twenty four. and become irrelevant as it happens. So yeah.

Um, so before we dive deep into about how does it happen and your backstory of all stuff, uh, people want to know what super memory actually is. Uh, okay, so for people who don't get it, what actually it is about and what the simplest way to explain what it does. Yeah.

Yeah, so here's the thing. You know, like we all talk to AI chatbots and many of these AI chatbots, like not many, only a few of these AI chatbots have memory like ChatGPT. But there's thousands and thousands of other chatbots in all sorts of verticals that would be a better experience with memory that don't have memory. And it's really hard to build something like this because not everyone has the resources like OpenAI or one of the big model providers does. So I'm kind of building a way for developers to add memory to their own app or thing that they're building. Now the question becomes, why wouldn't a model provider do that? Yep. And the answer to that question is memory is a stateful thing, which means it needs to be closer to the user, closer to the developer. It cannot be locked into one of the model providers. If you are locked into a model provider just because it has all your memory, tomorrow when there's a next best model, you will not be able to switch to it. So that's kind of like the core principle on which we build on. We also have a consumer app that kind of just uses this memory engine that we build and gives consumers a way to have a memory across all their apps and clients and everything that they do on the internet. which is more of a demo of SuperMemory, uh, as of right now. And the enterprise offering for businesses to build things is the main product.

Oh, pretty insane, uh, how you're differentiating with the current model providers. And in this space, there are many such memory enterprises working along. You have got, uh, appearances with them as well. I'll get back to this question, uh, again. But, uh, one of the things I would like to understand here is Uh, how are people using SuperMemory right now? What's the coolest or weirdest use case you have seen, uh, so far?

Um, there's like a very— there's some amazing use cases, uh, everywhere between companies building it to just developers, like indie developers building things. Um, I guess the, the coolest company, or like the most infamous company using it, is Clue.ly. Cool. Um, um, which is pretty cool because they're all around the news all the time, and I'm in their office sometimes, and it's pretty crazy. Um, but then there's like also general purpose agents, there's chatbots, there's a few AI girlfriends, uh, there's customer support agents, um, there's email clients, there's browsers, AI browsers, there's all sorts of things that use super memory today.

Can you highlight some of the use cases? I mean, okay, we talk about memory, but what Clouli is actually wanting to recall and in some other use cases as well. Yeah.

So let's say you add, let me give an example of Rube, which is one of my other customers.

And Rube is this, I guess.

Yes, by Composeo. It is this MCP that can do anything for you. Yeah. And if, if it has all the tools, it also has to kind of slowly understand the user, right? Because one day I will come and tell Roop that, dude, next time send all my work emails using dravya@supermemory.com. And when the next time Roop MCP is called, and I send— ask it to send an email and it's a work-related email, it'll use that email to send the email. So you can do like, you can extrapolate this to a lot of use cases. Like let's say I have like a travel agent and I tell the travel agent that, you know, like I am a CEO and I prefer window seats, et cetera, et cetera. It would know my budget, it would know that who I am as a person, and it would kind of customize. Its kind of options based on who I am and also book those things without me explicitly saying it. Or let's say you had like a general purpose agent, um, like Rube itself is kind of a general purpose agent. You ask something like, um, where should I go for dinner? Um, it knows that you're vegetarian, it knows, you know, like your preferences, like you're from India originally, so it will suggest places that are also better. So AI does not have this personalization layer right now, and we are adding this personalization layer, uh, to AI bit by bit.

That's pretty insane. And I guess down the line, this is gonna be the way. I mean, if model is having the context about what you think, what's your persona is, and creating a memo of it. So I mean, wide, wide use cases, of course. Uh, let's get back to your, uh, I mean, early days. Uh, you grew up in Mumbai, right? What were those early days like? The first time you got a laptop, what were you building? How's that been?

Yes, I am one of the lucky few to be born before, to be learning to code before the advent of AI. So I learned to code in the COVID era, like right at the starting of it. And then I left for a little bit and I was doing a lot of gaming. I was actually playing PUBG and Minecraft and all of that. And then I came back to coding, this time doing Python automations. And the reason I kind of love to code, like half of it was because, okay, I could see and I had something interesting to do, but the other half was because of the community aspect of it. Like I was always building these kind of semi-social tools for my friends, for people that I knew, and that really worked out well. Like my first product was a Discord bot. And that Discord bot had AI before the ChatGPT API existed, which is one of my big flexes. But essentially I used some other API. But the cool part about the Discord bot was it was like a general purpose bot, like it can do anything for you. And people loved it and people invited it to a bunch of servers. Like it was in thousands of servers at some point, used by hundreds of thousands of people. And it was my first successful product. So that's when I was like, damn, this is amazing. I should get into coding.

Oh, really nice. And how you're talking about this story, I remember, uh, one of the stuff you posted recently. I mean, when you were building, uh, actually the Twitter screenshot tool, uh, the one Hyperfully bought, uh, that was wild. How did that even happen?

Yes. So, I mean, uh, that has like a little bit of a backstory to it as well, because what happened was I started like the Discord bot thing. Um, that became like because the Discord bot used to be running on the corner of my room at all times. So then my friend reached out and he was like, dude, I will give you a server and we can put it on a server. And then we accidentally started a business out of that. So we were giving Discord bots out for hosting, and that's how I made like my first dollar on the internet. Um, we had like 8,000 customers and stuff. It was pretty big. That also got acquired. But on the same servers, I had this small project running which, um, was essentially an automation to keep an Instagram account running that I had. Uh, and the Instagram account literally just cross-posted tweets to Instagram. That's all it did.

Nice.

But to cross-post it, um, you first have to take a screenshot of it. And then I would actually just manually do that, and then I would put it in the automation and it It was like, it was not even an automation at that point because I used to do it all myself. I was like, there's got to be more people who are doing this. So I made a Twitter bot that can generate images of these tweets, and then later on I made another platform on which you can do this cross-posting stuff and stuff, uh, and a lot, a lot of things like that. Uh, I had a co-founder for that as well, uh, and I would like to mention him. His name is Om Shah. Uh, he's also in the US now, which is great. And we both did like— the timeline for all of this is like 1 week, by the way. So it was kind of insane.

Um, uh, what I get it like, the project you're kind of building is very practical. Uh, you know, you use Instagram, you use Twitter. If you're using it extensively, you find a use case here. If somehow you can connect both these 2, if this is some use case which can come out of it which is Uh, efficient or, uh, useful for anyone. So I mean, is that sense was, uh, something which people using is using extensively, or it was like picking up a niche and then starting it? Because what I get is it's just, uh, one use case, like one particular type of use case, like social media as general, uh, which people gonna use, uh, extensively.

Yeah, I was not smart enough to think about any of this, man. I was literally just building whatever I feel would do something. I did not earn a lot of money back then, at least before the acquisition. In fact, the Twitter bot was completely free. It had 100,000 followers at some point, but I earned $0. I tried a bunch of things, like I tried to do like advertisement posts and mentions and all sorts of things, but I was like, this will— this is not it. So there's no way, like, there's no reason for me to earn like a penny from this when I can go do bigger things in life. So yeah,

apparently the best thing comes out, which is solving for masses, and I guess this is what paid for it.

Yeah, and I also, like, I, I did not come from an extremely privileged background, but I came from a family that does business, and it's like you go big or you go home, you know, like Nothing in between. So that was kind of the mindset.

Some, I guess, Gujarati mindset that you keep. You have got the Gujarati background, I guess?

Yeah, yeah, I'm Gujarati.

Oh, nice. Um, that makes sense. Okay, uh, people talk about— I mean, there was some lot of news coming up. You left IIT, you dropped out IIT, you got into IIT. So I just want to, uh, ask what was the clear IIT story? You were on the IIT grind too, no? So what made you just step off that track and leave for the US? But before that, what's the whole story? What's the whole true story behind that? And the IIT story? What is the chicken media?

And I like this entire like Indian news, you know, push has been really funny to me because I I never got into IIT. I never dropped out of IIT. What actually happened is for the first four or five months after my tenth grade. So I did not give my 10th exam because COVID was going on, and then so I had kind of like lost touch with studying in a way. And then in 11th grade, I was like, okay, now I start studying, I'll go to IIT. And then I stopped studying because coding was much more fun, and I was already doing things that were pretty interesting. So and then the acquisition happened somewhere at the end of 11th grade. And I was like, dude, now why do I even need to go to IIT? I see all these IIT salaries and I'm getting it right now. Like, what's the point? So I just decided to kind of come to the US to live the American dream.

And I guess it's been fruitful so far. Of course. Okay, let's get back to your 40-week grind story. I read you built something new every week for like 40 weeks. Uh, what was driving that, the boredom, curiosity, or chaos? Uh, and did you actually ship all of them, or was it more about keeping your brain moving and figuring out more ideas to dive into?

Yes. So during the brief moment of like after getting my Twitter bot acquired to HypeFury, um, after that I was working at HypeFury for like 5 months or so, and it was like a job Um, and it was like not very nice. I mean, I was earning a shit ton of money, but I did not enjoy it. So I left, and I also realized that I am literally illiterate right now. Like, I literally don't know anything. I did not do my 10th, I did not do my 11th, I did not do my 12th. So I need to do things so that I know everything. So I ended up like, um, you know, deciding that if I'm going to do engineering, I'll do it well. And so I started like building all sorts of random things everywhere between like databases to consumer apps to like all sorts of things basically. Even like AI, of course. And I like one of my projects, and I shipped almost all of them actually on Instagram. You can actually Like, still just scroll through my—

your Instagram is a full repository of the work you have done. Of course, I have— I've just gone through, uh, just, just scrolled.

I did not get time to, you know, make a video for all of them, but most of them have something. Oh, and some of them are also on my YouTube. Many of them were actually for hackathons, some of which I won and some of which I lost. Um, like, 3 of them were actually at hackathons, uh, during that 40-week thing. Uh, one was CalHacks where I came first. Uh, one was, I think, Hack Princeton. Um, and the third was TreeHacks, I think. Um, but yeah, like that was at Stanford. So I was doing all these things and towards the end of those hackathons, uh, one of the more interesting projects for me was like, uh, was Super Memory. Because I was collecting all this knowledge. It was called Any Context because I was collecting all this knowledge from all these places in the world for myself, and I wanted to build something that I can use to manage my knowledge. Like to essentially, you know, I'm I have so much things that I want to learn, so many things that I have questions about that I've already read. Where do I ask them? So I essentially made like a way to put everything in one place and chart over it. Um, and then I made it open source and it blew up. So that was kind of the start of Super Memory.

Pretty insane. Uh, what was your learning strategy in these 40 weeks? I mean, uh, do you want to mention some resources, you know, figuring out? Because you were also starting very new, and you— since you said like you are not a grad, I mean, college is not a necessity anymore, but still, what's your Approach to learning? I mean, learning on the fly when you're building a project, or, or something beyond it?

I think, like, I genuinely believe in picking a direction and going as deep as possible in it. Um, like, like deeper than you would think you can go. Um, I did not do it for all my projects, of course. Um, some of them I went deep into the product aspect, like making it a good product. Like one was something called lumphao.tech, which was a meme thing. But the product was really difficult to nail because it has like a point system and it has like a notification system and bunch of things. On the other hand, like basically what I learned was, you know, I have a problem, I have like a few places I can ask them on Discord, like some communities and stuff, and I have the internet. So I would just, you know, mostly just read, uh, things every single day. I would like search things up and then dive, like, you know, follow all the links and then read as much as possible. Because like in the end, my thing was like, I have to get knowledge, um, and also win this hackathon, of course, because I was always participating in some hackathon.

Insane. So it was actually, yeah.

A few YouTube videos. To summarize, it was a few YouTube videos at the start, and then after that it was mostly reading stuff. But I still watch YouTube videos to learn things.

Do you want to mention some of them which really turned out very well for you?

It's like always like very random. Like for example, today if I was learning AI SDK or TypeScript, I would go, uh, watch Matt Pocock. Um, if I was, uh, learning Effect TS, I would watch Ethan Nizer. Um, like if I was doing Postgres, I would, like, if I was learning Vercel, I would go to Lee Robinson. It's like, there's like so many, you know, uh, small communities that you can be a part of and, you know, take advantage of them. Uh, I always feel the broader communities of like the tutorial mindset people are usually not the ones that are very fruitful. Like those tutorials are not useful whatsoever.

Totally makes sense. I mean, learning is— I mean, in the today's era, everything is available on internet and everyone can— if figuring out a product, learning can be done on the fly for sure. Yeah. You talked about AnyContext. I want to learn more about how the switch happened to SuperMemory. So somewhere in the mess came AnyContext, chatting with the bookmarks. Was that built in one sitting or something you kept refining? And when did you realize, wait, this might be bigger, this might be how AI remembers? Like, when is it even called Super Memory back then, or did that come after you started pitching to people? Yeah,

yeah, man, I have like a crazy story here. So what happened was when I built the first version of AnyContext, one of my other random products called Dumb.place, uh, it got attacked, uh, and it was getting like Millions and millions of requests.

And you should list all your products. You should tweet about it. What are all the products you have built in the channel?

Yeah, I had like a list of all my products. And by the way, I also have you know a list of failed things. Like I have a huge thread of everything that failed during that time.

That people should know before super memory what had happened in the in the past and how what worked and what didn't work. Yeah, go ahead.

Exactly. So actually, let me just share my screen and show you the list of failed things that I did. Um, it's like literally open right here, but okay. Like, I was building SpaceBot, uh, it had like hundreds of thousands of users. I did not— I was not able to monetize it. And then I was building EpicHost. I had, uh, like, you know, customers, and then the team fucked up. Um, and then, you know, I sold EpicHost, uh, at the right moment. I created an API that few people used. Then I made this thing called Audium, which was essentially like a Spotify for— What is that? It was a Twitter for audio. Um, that also did not work. Then I made like a developer tool. That also did not work. And then I made this kind of chart app. This one is really funny because it's one of my first like web apps and it was like terrible. That also obviously did not work. But for this, I had to learn a new language called Golang in like a night. That was the highlight of that. And this is when I made a Twitter bot and I kind of had this thing of like Dude, we need to get the Twitter bot to 1,000 followers in 30 days. And at this time, I myself had 21 followers.

What the bot does actually?

So I can just click on it. Um, so you just, you know, write poet this under any tweet and it gives you an image, uh, in this format that you can see.

Oh, got it.

That's it.

Nice.

Very, very simple. And then Twitter Decided to shut down their API and give me a bill of like $43,000 a month. So I went and sold this company as well. Then I launched lumpaw.tech. No one used it. It was a very community-oriented thing. I could not build that community. Then I built discord.pics, which was a Discord, which was a hosting platform, but on Discord. Like it, it was you can store your stuff on Discord. Like I reverse engineered it, and then I built like this thing called Siri GPT, and then bunch of other things like that. None of them were good ideas, basically.

Okay, so this was the whole story behind how any context has happened. Yes, and how many super memory has happened.

So all of this gave me like an idea of okay, I should just build for myself. if I'm building in consumer. And the best, like what happened was when I open sourced, I was open sourcing everything. And so, but when I open sourced AnyContext, it started getting a lot of traffic because I had this viral tweet that said like, this is probably worth $1 million because I've worked a lot on it. So, but you can just take it for free. And dumb.place was getting attacked at the same time. So I had to reverse engineer Cloudflare, um, to save myself. And then I accidentally got a job at Cloudflare.

Yeah, I totally guessed that this was the pivot.

And then at Cloudflare, I, um, realized like infrastructure at scale is actually not that hard, um, if, if done right, like if the engineering is right. So I ended up like Um, kind of re-architecting AnyContext to the Cloudflare ecosystem, um, and make it in a way that it can run in $5 a month because that's the money I had. Uh, I was like, literally I had run out of all my money from acquisitions by this time. I, I had borrowed $1,000 from my friend because I was in the US, okay?

Yeah.

I had borrowed $1,000 from my friend and $800 from my girlfriend to just live. And I had this other projects going on. It was an insane time. So that happened. And then this random guy comes to me and he's like, "Dude, I have the domain supermemory.ai. Do you just want it?" And I'm like, "Sure, that's a great name." And I'm like, "What do you want in return?" So he says, "I am one of the lead developers of React Native at Meta." And all I want from you, because I really think this is a good product and you are a good builder and this will turn into a company. If you raise for Super Memory, I want to invest.

Nice.

And then two years later, like one and a half year later, I ended up raising and now he is actually in the cap table. But that's a pretty insane story, and that's how it came to Super Memory. I

mean, the best relations. In terms of investing or whatever, the best relationship you get in—I mean, what is done when you're building something? You you are in chat with someone; they're understanding the stuff. I mean, you are not pitching very new, but you are—they have been the part of journey so far down the line.

Exactly. Exactly.

I saw your tweet, like you posted about how your girlfriend supported throughout your journey. Can you talk more about this? Like people have very different opponents on this. How's that? How has this been for you? And I mean, I feel pride too, but I'm just want to listen you how how you how you think about it. Of course,

yes. So essentially, it's really difficult to have a girlfriend, and it's usually not recommended for founders to have one, and I understand why. But there's like what happened was I. Came into ASU where I studied, with this mindset of I want to build a lot, and so like no girls, no parties, no chill. You know, I just want to build. So I, I was just building, and she saw my failure tweet, and then she con she was she was walking past me, and she came, and she was like, "Dude, I saw your tweet. You want to go for a walk?" So I went for a walk and I explained everything to her.

This is such a movie tale. Really? Yeah.

So like the story starts there. So I mean, of course, everything else she knows that I'm a builder and she has always been supportive of that.

Oh, so so she has been seeing from your lows and now up to the mark. I mean, such a crazy story for anyone for anyone listening. I mean, this is this is this is so damn good. Um, you also talk about solding companies, you know. So how you define how you can pitch to, uh, how you can, uh, address this? Like how you're gonna sold something you have built? Because, because in India, anywhere in the world, people build stuff, cool stuff, but they— but that project just get limited to resume. They, they don't get a chance to pitch to someone to actually think about solving that stuff, selling that stuff. So yeah. Wanna hear your thoughts on this.

I think if you build a company or a project from the perspective of one day I'll sell it, it will never be sold. In fact, both of my acquisitions happened out of desperation, as I mentioned before. And they were less of acquisitions, they were also very acquihire kind of oriented. Either that or I had a lot of users. Um, in both. So like, you have to get to that critical mass where it would make sense for the buyer to get anything, right? And

after that, which is actually the most, uh, the GitHub stars, the people using it and all stuff.

Yeah, the GitHub stars is fine. No one cares about that actually. It's just that it should be loved by many people. Like, making anything that's loved by even 10 people in the world is extremely difficult. And if you can do that, then it's probably going to be a good thing.

Makes sense. Totally. Um, also wanna talk about your previous work. I mean, you worked at Memzero, you worked at Cloudflare, and then suddenly dropped out and, uh, I mean, get out of and working for SuperMemory full-time. So how the journey has been at Memzero? And I, I remember Memzero is solving the same memory problem. So what's the differentiating factor here between Supamemory and Memzero, and what you learned from that experience?

Yes, so it was a very interesting time because, uh, I joined Memzero right around the time when I was at Cloudflare. And at Cloudflare, I had done like a very like interesting research work. Um, I actually filed a patent there, um, and I was given a lot of ownership and all sorts of stuff. And the reason I joined the startup was because Of course, it's like an early-stage startup. I'm the first hire ever. I will be able to do incredible stuff. And I've known this guy for a while. I've been in the memory space for a while because Super Memory existed before I joined. So I was like, this will be cool. And then I see, like, first of all, they are not— like, I did not quite believe in a lot of their engineering practices and research practices. Um, they were doing their research for the sake of it, uh, manufacturing numbers and stuff. And this is a very well-known thing. This is not conspiracy. Yeah. Um, if you see, I— in this space, there's 4 companies, uh, one being MemZero, and all 3 collectively agree that the research is not right. Um, and the entire internet also pretty much knows. So I was like— and I, I was there when they were thinking about it. I was like, this is not what I want. Also, I did not have time because I was in university. So I was like, I would rather just focus on my own thing than be at a startup where I don't feel good. So I continued with Super Memory. And as for the differences, well, I would say the biggest difference is like, there's 2 big differences. One is latency and one is quality. And then the third one is features. Um, in terms of latency, we are about like 8 to 10 times faster in most cases. Um, because I come from an infrastructure and research background now, um, so it's like— and I also have partnered with Google, Cloudflare, PlanetScale, and a bunch of other companies to keep this low. Um, even like Stripe and Autumn, like every kind of subprocessor in the startup like in this entire pipeline is actually working with us very closely. And the second one is performance. So the way we— so the way like others measure memory is not right, in my opinion. You can only have good ways to measure how good your system is if you have had an experience with real users using that system, which we had for years because we had the consumer app, which is where we started. So what happened is I had this consumer app, I figured out the $5 strategy, you know, like of scaling this thing for almost no money. And then I kind of became a consultant for a lot of startups. So I had a lot of, you know, data on my own startup. like on my own kind of project that I was running and these 20 different companies and how their users use their apps. So I essentially made my own like little benchmark and everything. But it boils down to 3 things. First is temporal data, which is like how do you think about updates and things that change? Second is forgetfulness. When do you forget it? Why do you forget it? And third is inferences, like two random pieces of information should connect together into new information. That's how kind of the that's how the brain thinks about things. So I modeled super memory on top of the brain instead of just having it like as a CRUD store, like just a database store. So that's the story. So we are like faster, better, cheaper, quite literally than. the alternatives.

Oh, makes sense. I mean, that's the— I mean, that's the whole point, uh, people gonna go for it, uh, enterprises and users, everyone. Uh, also curious to know more about the Cloudflare story and how did it happen. I mean, yes, I mean, the CTO has shared some of the very good, uh, tweets about you. Uh, how did it happen and how you get, uh, the CTO on board in this race? Um,

When I joined Cloudflare, my— the guy who was supposed to be my manager left Cloudflare to do a startup.

Oh.

So I was this guy without any manager, and Dane had hired me personally. Like, he sent me a DM on Twitter to hire me, so he knew that I'm, I'm a builder without a manager. So I kind of became like Dane's, like, you know, little guy who goes around in different teams and fixes things. And I was given like a lot of ownership and, you know, a lot of permission to learn and talk to people because I was still dumb, you know, like I did not know a lot. In fact, Dane would sit next to me and watch me code and he would be like, this is terrible code, but go ahead, we will fix it on Monday. So, and in the office we would sit together as well. So he has been, you know, like very vital in this journey. And Cloudflare as a company, like there's everyone I know in Cloudflare, like Sunil Pai, um, who is 3.1, who is like one of the legendary, you know, programmers. But also like the unsung heroes, like all the PMs, uh, all the engineers, they still help me. Like they are in my messages Helping me, which is great. So yeah, like I got it just by building and hacking stuff, and then Dane accidentally became my manager, and then here I am.

Insane! I mean that that's what what what a individual who's working for a company would demand. And if if I mean the I guess it totally boils down to how better you're in a relation with the the good people you're surrounding with. Um, I wanna— let's talk more about the, the Raise story. I'm curious, how does a 20-year-old even get people like Jeff Dean, Josh Browder to back them? Uh, were they already following your stuff, or did you cold DM your way in? How did it happen?

So I was doing BuildSpace. So Super Memory was a part of BuildSpace, and we also won BuildSpace. um, which is like this thing for builders in 2024. It shut down. Um, but right after BuildSpace, after I won, I was like immensely viral. It's not immensely viral like in today's terms, but I had like 300K views or around that number. So I was getting DMs from a lot of people. Like, there's this one founder who flew me out to New York and gave me like full like VIP service in business class. And he was like, Dude, come work for me. Drop out. Do like I will. You know, I will take care of everything. And then there were other people who were DMing me. Like whenever you raise, let me know. And then in January, A16Z reached out to me, and they were like, "We wanna you know put in money." And I also got the YC Summer Fellows Grant. Um, which is for individuals who want to do research in the summer. And I was like, man, like, this is kind of crazy. Like, I'm not even serious about what I'm doing. I'm doing it part-time while having a job, uh, and I'm still, you know, able to do this. So I should do this seriously. So I started doing it seriously. I did not take the a16z offer. Um, I told YC to not give me the grant because I was not a student anymore. I was going to You know, fundraise and build it into a startup. So YC was like, "Dude, why not you just raise money from us?" And then Julian Weiser, who is funnily enough my neighbor right now, he's like right upstairs. He can probably hear us. But he called me and he was like, "Dude, I know you did ODF, which is one of the programs I did in San Francisco." He's like, "If you're starting a company." If you had to do it solo, like, are you looking for a co-founder? I'm like, I'm looking for a co-founder, but it's only because others like investors asked me to. He was like, I'm starting, I'm thinking about starting this house. It's called the Solo Founders House, and you'll be the first one to join. Like, if you say yes, I'll do it, and you'll have a place for someone, and you'll build a startup. Nice. And then I said yes, and then that happened. But then the problem was I don't have a visa to drop out of university. So Julian also helped me with that, and that was great. Like, I did not have money, so he sponsored the visa as well. Like, he put in money as an investor in Super Memory, and then Super Memory sponsored my visa. So it was pretty good. So he was also the first investor before A16Z and before all of these. And he also was helping me along this journey as well. And then I DM'd back the people who had told me like, dude, I will invest once you raise. I told him I'm raising now. And that started this chain reaction all around SF that Dhravya is raising. Because like Dane had, like, and people in Cloudflare had already seen and spoken about me. That this is, you know, when he raises, it will be nice. And that led to like VCs just reaching out and like a lot of cold, not cold, but like warm connections being made, like intros being made, and then the raise happened. And then the Jeff Dean story is, that was a little bit different. So, uh, during the Build Space time when SuperMemory was growing, um, Logan Kilpatrick, who is now one of the product leads at Google DeepMind, um, that's true, he messaged me like, do you have any interest in joining Google? And I said, yes, of course. Um, that did not work because I would, I would not be able to join full-time because back then I was still in college. Um, and then I also had an offer from like xAI, and then I got one from Anthropic. Um, and then I reached out to all these people. I'm like, dude, you were going to hire me, why don't you put a check? Uh, and everyone said yes. Um, and then that's how I got like Jeff Dean as well to put in a check.

That's totally insane, dude. I mean, I mean, this is one of the— I mean, of course your dream come true to have people like Jeff Dean and the folks to have you in, of course. Also curious about your YC story. I mean, YC invited you to write. Why you'd skip it? What was your thought process around this?

Yes, so right after I said no—

Millions get into YC because you get the crazy network and it opens the door to various enterprises. How did it happen?

Yes, so right after I said no to the Summer Fellows grant thing, uh, and I told them I'm building a company, Gauri Tan called me and he was like, uh, you should just do YC. Um, and the reason I did not do it was because it was kind of too late at that time. I already had offers at like, like very high valuations as compared to YC, and I had also been at Like a few YC companies, and I knew that this is not like this is not that crazy like as people think. A company can be built nevertheless. It is about the support and stuff, and if I was starting a new thing, I would still do YC. But since I was already kind of in the game for years, and I knew exactly what I'm going to do. I decided to take like the bigger, you know, institutional funds.

Makes sense. Totally. That justifies. Let's talk about more about your O-1 visa story. I mean, that's like the you're officially extraordinary in America. I mean, it's a really great track for you. How did that feel? And was it validation or just one more form you had to fill? And how did it happen? I mean, what are the also if you can mention what are the checklist to have? Uh, if you— if someone is looking for O-1 visa, how that process is,

right? So, um, when I originally wanted my O-1 visa, which was last summer, um, I— like, all the lawyers I would reach out to would say no because I did not have that good of a profile. Um, but after doing the patent and after, you know, getting all those offers and all of that Uh, before I, before I fundraised, um, I was like, I would have to get the O-1 visa. So I spoke to Julian, and Julian told me things that I did not know about myself in a way. Uh, like, maybe I can also put this in. So I put like media recognition, which I did not know that I had. Um, I put an original contribution from my open source work because at the time Super Memory was one of the biggest, like large, fastest growing open source projects in the world. And okay, so high pay at Cloudflare and before in Mumbai as well. Distinguished role in a distinguished organization, which is also at Cloudflare. And then patent, which is research. I had membership at ODF, I had judging at Hack Illinois, and I had award from BuildSpace and others.

I mean, that justifies— I mean, you get all the fillers to fill in for sure. Let's move ahead. And I've got this question from Vizor itself. What's your experience being a solo founder after having co-founded successful things in the past?

Yes. So I had a lot of co-founders before. For my original startup, EpicHost, the Discord bot hosting thing that I mentioned, that actually failed because of a co-founder breakup. And it was really bad because we lost everything in a second. EpicHost could actually be a big company, like a Vercel-like company. Um, but I was like, dude, this is like stupidity. We were 15-year-olds in, in our like houses around the world, like one in Europe, one in India, one in Australia, one in the US, um, building something that's amazing, that's used by thousands of people, um, and we, we stopped the company because of some feud. So I did not want that. Um, secondly, I did not want the thing of co-founders of convenience, which I felt I had before in my second startup and after that as well. Also, sometimes it's hard to align people on the vision. And even if they're aligned on the vision, they might not understand tech enough to be doing infrastructure, especially AI infrastructure. Obviously they don't have to code, but they have to understand how it works. So all of these criteria, and then I actually trialed with like almost 8 different people for Super Memory. Um, I tried getting a co-founder, and I, I will still be happy to find one. Um, and every single time there was something off that I was like, long term, I cannot see myself building like a billion-dollar company with this person. Um, And maybe that's my own trait, maybe that's like, I don't know. But my basic thing is I want to build a business. I will do anything that's good for the business, whether that's getting a co-founder or not. Um, but I don't want the business to do worse because it has 2 people running it. Um, and so far I have seen it's, it's going pretty good because I was living in a solo founder's house and because There's a lot of solo founders living with you. The concept is that you guys are solo together, so all the things, all the emotional aspects, all the difficult parts of running a startup, which is when co-founders are actually needed, you just have other solo founders to talk to, and they kind of become like, oh wow, I now have six co-founders in that sense, which is great.

I mean, how it how easy it is to say, I mean. It's difficult when you are running on the process. You spend time with them, and you are running out of this decision to take off it. I mean, it's totally incredible of you to. I mean, the vision is very clear to to grow up the business, to grow up the company, and totally makes sense to have some people off for sure. Let's get more, and let's interested in your hiring for super memory as well. So what What's the work culture at Super Memory? How how how's your how's the routine of the people working in the company? Just give me a sec. How's the routine of the people in the company and how do you hire? What you're looking up to? Yeah.

Um, so the routine of the people in the company right now since the last one month has been. Wake up at 7 AM, at least for me. Wake up at 7 AM, freshen up, do your own things like talk to parents, whatever. And then around 9 AM, we take the meeting, like the team meeting. And then some people come to the office in San Francisco, which is right above where I live. And we then work. An entire day for 12 or 14 hours, um, till like 11 PM or whatever, and then everyone goes home. Like right now it's 11:27 PM and I have more work to do after this. So it's basically like, it's a startup culture, like that is how it works. How I hire is the person should dive as deep as possible into things, should be able to provide value, should be, should be extremely communicative and responsible and sincere and disciplined. Um, and you all need to

work— Ownership. Uh, you take ownership.

Yeah, exactly. It comes down to ownership. Like, if I dropped you in some task, you know, you should be able to do it. If you're not able to do it, then you should be able to ask for help. Um, and, um, the last one is, of course, you should fit in the culture. You should be fun to be with. You should be a good person. That is like table stakes. So that's kind of what we look for. Things have been like up and down. We have fired people. I have fired like many people, by the way, at Super Memory. My concept is I hire fast and fire fast. And even if that means like temporary, you know, problems, like if someone wrote the entire infrastructure and I had to fire them, that's fine. that will be a bad thing right now, but then like in the future of the company, it's going to be fine. Now I found the best people are people I had met years ago. One of them being Mahesh, who is leading like most of our product and a lot of our engineering right now. And he was an open source contributor in the open source super memory 2 years ago when I open sourced, like when I released any content.

Relationship you have met so far, and I mean the fruitful ones.

So I ended up hiring like a few people from, you know, years ago. Um, one of them is Soham who leads research now, and Souptik who I met online 3 years ago because I liked his designs. Um, so yeah, I mean it's kind of insane, like I have hired all of them through something I knew from before because that's what the early team should be anyways.

Makes sense. And, uh, what's the headcount and are the people remote or everyone is in office in SF?

Um, we have 2 in research, uh, that's Soham and Sriram, 2 in engineering which is me and Mahesh, um, and 2 others. So one guy does like— SoupTech does Uh, design, and Naman does content and docs and blogs and all of that,

all in, in person at yourself.

Yeah, no, most of them are not in person actually. So 2 are in Arizona where I was studying, and that's where, that's where I met them. Um, 2 of us are here, me and Mahesh, um, and 2 are in India. One is actually in Singapore. So one in India, one in Singapore. So it's a remote team. Mostly, but I fly back and forth all the time. Like I'm going to Tempe tomorrow because I obviously want to. I also do a large chunk of the research and I have to be everywhere at the same time.

Totally makes sense. Inside man. And next is, do you? I mean, what do you think about what the differentiating factor between if you have if you if you didn't move to U.S. If you happen in India. So what the difference you find in the ecosystem and, uh, what it means to you actually? Do you think you would have gotten the same level of the trust from Indian investors as you did in Asaf, or in, in every trajectory possible?

Yeah, I don't think people quite understand journeys and stories the way it works. Um, it's not like I just dropped into a place and I asked for investment. It's through years and years of like knowing people and like working hard. And, you know, the random guy I met 3 years ago at an open source conference in SF now is my investor, and he led to a lot of other investors. And that only happens when you allow serendipity to happen. Like, for me, it's not about India versus US, it's the Bay Area versus everywhere in the world. You know, obviously this is where the internet is made and this is where the people who want to put in money are, both customers, investors, and engineers. So like everyone. So I would be here too because this is where everything's done.

Makes sense, totally. Also, as a first-time founder, which do you prioritize more? I mean, the vision or the profit? I mean, apparently you're going for it, so you wanna have to have maintaining ARR and all, especially with the investors now on board.

Yeah, um, right now the main focus is actually not profit, contrary to popular belief, um, because that's not how venture-funded startups are made, um. In the early days, all that matters is you find something that people really, really want and you can get it to as many people as possible. I believe people want memory. Now my next thing would be to be like the only choice to do memory in the world. And after that happens, I am able to do everything else, you know, like I'm able to make money and stuff like that. I obviously have like a very good like MRR and ARR and all of that right now, especially for funded companies, especially for companies that started months ago. I'm not at a bad position, but that is not my focus basically.

Got it. Would you consider going back to college again, maybe for master's or other degrees? I mean, to dive more into the research area now?

I thought about it, and I would rather go to one of the frontier labs instead of universities if I really had to do it. I am very involved with my friends at ASU, so it's not like I'm missing out on the college life or anything. I still have a college life. The culture in SF is also pretty much like a college. Like, I can wake up, go to a friend's guy. Place and then we can go for an ice cream and stuff. So no, I will not go back to college.

Makes sense totally. I mean, I mean, I interviewed Kalomaz as well and he was not looking forward to that because the in the ecosystem you are in, I mean, you are the best people around you and and and is is the thing you can actually get for sure. Um, also what the next big thing Super Memory is working around? Yeah.

Yes. So this is interesting. The next big thing for us, there's 2 big things, one for the developers using and enterprises using us and one for the consumers. For enterprises, we are releasing like a local-first version of Super Memory, which you can ship like on device on your customers' phones or laptops or browsers or whatever they have. And for consumers, we are building Uh, we are kind of revamping the app experience, um, and we are renaming it to Super Memory Nova. Um, so it's an agent that knows everything about you. So it's gonna be pretty interesting. Few weeks, months, um, weeks most probably because we move really fast. Um, but yeah, very excited for what's next.

Interesting, man. I guess, uh, I guess that's a wrap actually. We are pretty much end of the pod. And actually, from a kid in Mumbai building stuff to raising $3 million, giving AI the power to remember, this one is wild. And what I love most about your story is like every such pivot has a craziest story around this. And some of the stuff like obsession, ownership, the relationship you have met so far, and the learning approaches and Figuring out resources, what to do, and figure out a product of it and selling it actually. Totally insane. I guess it's not just a startup; it's a signal of where AI is headed. I mean, memory is one of the things for sure, from reasoning to remembering. And yeah, honestly, this convo felt like like not not like a podcast interview, but more like watching the next chapter. What's coming in and how one can build one of the coolest stuff being being social and being just being obsessed. Uh, big thanks to you for coming on and to everyone listening.

One, one thing I want to mention, um, is like the media and the people have kind of made this thing into, oh my God, this guy raised $3 million, he is making it in life, he has made it in life. But it's actually just the start. Like, I have not raised $3 million to sit on it. So I'm just getting started and there's a lot of cool things to come in the future.

Pretty interesting. And on one last note, if you want to say something to people who are just starting out with college and they're in high school and figuring out to— I mean, AI is the stuff which is very frontier of the tech right now. How should they dive deep into it? How can they be one of the— on the top 1%, you say? What should be the right hard approach to that? I mean, it's not— of course, it's not a roadmap, but it's sort of the self— how one can be obsessed with what they're doing. But with the current scenario, what you're saying, you are at the top of the tech right now. So what's your view on this?

My simple view. So funnily enough, my cousin just called me a few hours ago and he asked me the same question. And I'm like, do what you genuinely, genuinely feel interested in doing and you will dive very deep in and you will be interested in. Like if you're already not in engineering and you don't love engineering from the bottom of your heart. Doing AI, like in terms of like AI research or AI coding or applied AI, does not make sense because that's not the thing for you. There's, there's a million other people doing that. Do something that you are already like have taken a few steps towards, um, and excel at that because you can achieve excellence in anything if you just have to choose what you want to excel at. So that's what I would say.

Uh, I mean, yeah, totally makes sense. And I guess, yeah, that, that's all about this podcast. Thanks for tuning in, the advertisements alone. I mean, this is the first podcast, uh, first interview after the raise, and, uh, seemed very pumped up and looking forward to what the next— I mean, I guess, yeah, this is just a start. Uh, you're 20, you raised $3 million. I mean, it's gonna be exponential for sure, and I'm looking up to what more trajectory is gonna open up, not just around memory but other stuff as well. Uh, yeah, man, thank you, thank you, thank you so much for joining in.

Yeah, thank you so much for inviting.

Cool. Uh, yeah, that's all for this episode of Ground Zero. See you in the next one. Okay, bye.